Commit Graph

Select branches

Hide Pull Requests

abstractions_4

add_local_buffer

after_qol

algebraic_upat

all_bv_no_idx

all_sink

amd_isel

amd_refac

amd_smu_reset_patch

amd_sqtt

amd_uop

amd_uop_cleanups

assign_copy

assign_try_3

before2

better_spec

better_syn

bitcast_spec

broadcast_shape_2

bt_ei

buf_heur

bump_nv_2404

cache_reshape

call_inline

call_is_kernels

callify

cast_vec_as_cast

cat_mop

changes_from_rangify

check_clsetarg

check_dims

cifar_30

cifar_30_clean

clean_load

clean_spec

clone_tg

codegen_try_2

contig_on_after

continue_mat

deepsek

default_late_allreduce

delete_kernel_py

delete_ones

delete_slow_rangeify

delta_attention

delta_attention_scan

demote_op

disable_rules

dsp_search

dsp_search_merged

dtype_shape

earlier_gater

emu_cdna

emulate_ctx

end_all_ranges

endrange

er_prioity

execution_unit

failing_assign

fancy_usb

fast_llm

fix_fa_fwd_clang_22

fix_fuse

fix_llvm_cast

fix_range_merging_sd

fix_rangeify_tests

fix_rc

fix_sched_copy

fix_slice_store

fix_src_spec

fixup_fix_kernel

full_spec

fun_w_egraphs

gate-on-load-store

gate_rewrite

gemma4_gpt

gen_pdf_fast

gh-pages

gl_dims

gldims

glm_flash

grad_outerworld

grok_x86_novec

group_div_rules

hcq2_vram_copies

heur_4096

heur_group

hk_gemm

http_threading

hwtest_fixes

id_store

if_range_cleanups

ignore_oob

image_idx_xy

image_in_mc

image_no_vec

image_refactors_stack

index_to_slice

invalid_try_3

jinja_template

kernel4

kernel_is_call

kernelless

kimi_slop

kitten_matmul

late_Locals

late_add_load

late_allreduce

late_codegen

late_decomp_is_late

late_lower_idx

late_view

late_weakint

lds_double_buffering

less_wmma_arg

lil_dg_change

lil_rangeify_td

llama_trainer

llm_262k_minimal

llm_flash_attn

llm_qwen

llm_serve_fixes

load_view_late

local_cache

locals_have_device

loop_end_op

loop_is_rng

lowerer_hacks

mac_pytest

manual_kernel_3

master

mega_lowerer

min_new_lin

min_outer_range

minor_views

mixins7

mlperf_training_v4.1

mlperf_training_v5.0

mlperf_training_v6.0

modern_scan

more_apps

more_early_comps

more_folding_rules

more_kernelcount

more_pcode

more_rangeify_pm_tests

more_ren_clean

more_uop_gc_test

more_uop_mm

more_vmap

move_devec

move_dtype_decomps

move_expander

move_gates_to_load_store

move_mop_cleanup

move_schedule

move_simp_views

moveleftright

mt_typing

multi_range

multioutput

new_ck

new_devec

new_invalid_and_image

new_linearizer

new_shape

new_test

no_assign_2

no_buf_0

no_decimals

no_late_simplify_marg

no_merge_ends

no_merge_views

no_mstack_in_allreduce

no_one_ranges

no_sip_usbgpu

no_tsink_base

no_tuplize

no_user_ck_contig

no_vec_x86_4

no_weakint_variable

noop_index

novecimg

num_batches_tracked_shape

olmo3_support

one_compiler

one_hcopt

only_assign_buffers

only_loop_backends

only_reg_emu

only_reg_emu2

outer_range

outerworld_work

pad_invalid_work

pad_mixin

param_call_is_function

parse_sqtt_more

pci_iommu

port_nir

postrange_hcopts

precompile_backward

prefill_qwen_fast

print_special_ops

ps_clean

pyremu_try

python_prg

python_speed

q4k_support

qcom-openpilot-13ms

qcom_830

qcom_mmapeak

qcom_speed_codex

qol_string

qwen36_27b_amd_900

qwen36_27b_amd_minimal

qwen36_amd_bugfixes

qwen_mergable

r_cleanups

r_new_opt

ranged_call

rangeify_2

rangeify_const_folding

rdna3_vibes

rdna4_asm

rdna4_gemm

real_pyrender

recurse_sub

reg_coalese

relu_pattern

remove_assign

remove_cast_hack

remove_ck_contig

remove_define_var

remove_forced_re

remove_on_stack

remove_programspec

remove_pylint

remove_slow_tests

remove_vec_2

remove_vec_3

renderers_use_shapes_2

replace_if_with_range

reprocess_node_experiments

reshape_trait

resnet-ci

revert-11547-fix_mismatch_red

revert-11715-revert-11711-amd_llvm_tc_cast

revert-11809-no_ki_in_gpudims

revert-11811-better_div_nesting

revert-11876-memplan_vbig

revert-12104-simplify_valid_in_symbolic

revert-12308-bufferize_early

revert-12731-jitlink_init_check

revert-12813-after_cleanups

revert-13046-better_cleanup_arange_buffers

revert-13311-prof_link_sqtt

revert-13583-rocr_desc_2

revert-14734-tload

revert-17141-full_rewrite_to_sink

revert-17408-_broadcasted-weak

revert-9438-simple_matmul_np_rand

revert_wrong_ck_fix

rewrite_rangeify2

rtoposort

scan_assign

scan_op_8

sched_cache

sched_lin

sd_load_simple

self_type

set_testpath

shrink_in_render

simp_spec

simple_gidxs

simple_pool

simple_priority

simple_rangeify

simpler-sd-compile

simpler_end

simpler_fusion

simpler_image

simpler_postrange

simpler_reduce

slice_like_shrink

spec_3

spec_cleanup

spec_ren

speed_one_linear

split_buf_idx

split_decomp_pass

split_sub

stack_0

store_is_void

string_ranges

support_opts_in_contig

swizzle_in_kernel

sym_llm

sym_work

symbolic-chunk-split

szdiff_bot_test

tc_is_heur

tensor_graph_single_sink

test_amd_crash

test_fa

test_from_postopt

test_hgemm

test_unit

tiny_scan

tmr

tool_call_support

topoprop

ttomsa-new_linearizer

tuplegtuple

two_stage_remove

typed_checks

typing_work

unused_permute_arg

uop_prg

upd_test4

update_benchmark

update_benchmark_search

update_mlperf

user_failure

vecless_x86_glm

verify_sink_dims_clean

vf_ish

view_after_const

view_in_codegen

viz_cbroadcast

warp_fun

wgpu-f16-packed

working_postopt

x86_moves

x86_numel

v0.10.0

v0.10.1

v0.10.2

v0.10.3

v0.11.0

v0.12.0

v0.13.0

v0.4.0

v0.5.0

v0.6.0

v0.7.0

v0.8.0

v0.9.0

v0.9.1

v0.9.2

Mono Color

• 0f0627bc60 add mnist tutorial geohot 2024-04-25 16:08:32 +08:00
• d31e220cbf add mlperf-logging to setup.py mlperf (#4289) chenyuandGitHub 2024-04-24 23:34:34 -04:00
• 6b8a85939d fix lds size for amd (#4287) nimlgenandGitHub 2024-04-24 22:54:42 +03:00
• c11bad766d prepare mlperf submission (#4270) chenyuandGitHub 2024-04-24 13:19:31 -04:00
• c606a0ba6f Docs link fix (#4286) Szymon OżógandGitHub 2024-04-24 18:54:43 +02:00
• c1fbacb182 resnet benchmarks use DEFAULT_FLOAT=HALF (#4285) chenyuandGitHub 2024-04-24 12:10:57 -04:00
• 002a14088e Ptx store gate cast to bool (#4284) Szymon OżógandGitHub 2024-04-24 17:43:44 +02:00
• dbe3e1d548 or true fixes ci (#4283) George HotzandGitHub 2024-04-24 16:48:26 +04:00
• 53853e6d08 save the schedule graph in SAVE_SCHEDULE (#4248) qazalandGitHub 2024-04-24 12:08:51 +03:00
• acb32e1766 hotfix: PM4 supports timing geohot 2024-04-24 08:38:59 +00:00
• ad28fdecb1 si.inputs+outputs -> bufs (#4279) George HotzandGitHub 2024-04-24 11:12:34 +04:00
• 8401de9922 resnet benchmark return early in eval (#4278) chenyuandGitHub 2024-04-24 00:55:01 -04:00
• 38f97aa0fe rename rawbufs to bufs in ExecItem (#4274) George HotzandGitHub 2024-04-24 07:27:27 +04:00
• 60e3aa5cb1 more docs (#4271) George HotzandGitHub 2024-04-24 06:52:42 +04:00
• 6637ecc5fe use IGNORE_JIT_FIRST_BEAM to not BEAM in jit cnt=0 (#4269) chenyuandGitHub 2024-04-23 18:59:43 -04:00
• f3b4dff7c9 KFDProgram -> AMDProgram (#4268) nimlgenandGitHub 2024-04-24 00:29:50 +03:00
• 17328ded7d setitem no return value (#4266) geohotstanandGitHub 2024-04-24 04:28:14 +08:00
• 3a48773f1a BERT dataloader (#4252) Elias WahlandGitHub 2024-04-23 19:44:49 +02:00
• 69341144ba Wikipedia preprocessing script (#4229) Elias WahlandGitHub 2024-04-23 16:28:01 +02:00
• 759b4f41c3 few more KFD -> AMD (#4262) chenyuandGitHub 2024-04-23 10:15:37 -04:00
• 6c25f1abf7 Optimize ptx loops (#4263) Szymon OżógandGitHub 2024-04-23 10:20:14 +02:00
• 967638f0d5 update docs, remove corealize (#4264) George HotzandGitHub 2024-04-23 12:05:29 +04:00
• 9b7efa72ea hotfix: skip 0 line count files in sz.py geohot 2024-04-23 11:56:03 +04:00
• acf4ba5c9f method cache respects beam option (#4261) George HotzandGitHub 2024-04-23 09:00:41 +04:00
• 9a95781d51 renamed (#4260) George HotzandGitHub 2024-04-23 09:00:28 +04:00
• 2ae4f45272 WIP PM4 Support (#4110) 3 people 2024-04-23 08:31:27 +04:00
• 3f6c7ca8bf test: fix test_tensor_core_padded on CUDA and add to benchmarks (#4258) Francis LamandGitHub 2024-04-22 20:22:11 -07:00
• a90de3b574 search: add additional 7 factors to the action space (#4256) Francis LamandGitHub 2024-04-22 16:14:23 -07:00
• de2b1fb468 update adding_new_accelerators doc (#4255) chenyuandGitHub 2024-04-22 18:50:19 -04:00
• bbb0ad4800 wmma: widen TC usage in search by using PADTO on TC axes when possible (#4216) Francis LamandGitHub 2024-04-22 13:50:31 -07:00
• 9e53d6cffa hotfix: 8000 lines geohot 2024-04-22 20:58:16 +04:00
• e6227bdb15 nv driver (#4044) 3 people 2024-04-22 18:50:20 +03:00
• 77a3780005 assert reduce recompute (#4250) qazalandGitHub 2024-04-22 16:12:39 +03:00
• a9bc7c1c49 unify assign tests (#4247) qazalandGitHub 2024-04-22 11:01:15 +03:00
• 37f8be6450 resnet print epoch ops and mem in benchmark (#4244) chenyuandGitHub 2024-04-21 18:32:31 -04:00
• 7bc862767c Improves error message when CUDA module fails to load. (#4243) Micah ZoltuandGitHub 2024-04-21 23:10:14 +08:00
• 4c99d49c4d some docstrings (#4201) 3 people 2024-04-21 08:34:08 -04:00
• 30fc1ad415 remove TODO: remove explicit dtypes after broadcast fix in stable_diffusion (#4241) chenyuandGitHub 2024-04-21 00:31:24 -04:00
• a1940ced77 remove the assign hack in whisper (#4240) chenyuandGitHub 2024-04-20 23:56:44 -04:00
• 3f126c7664 fix examples vits / converstion.py (#4239) chenyuandGitHub 2024-04-20 23:29:12 -04:00
• 31c9d9a228 fix test_linearizer tc opt tests for bf16 (#4237) chenyuandGitHub 2024-04-20 11:51:50 -04:00
• f1d9d0a151 cleanup external_test_opt (#4234) chenyuandGitHub 2024-04-20 04:00:08 -04:00
• dc4b1af09c more realistic edge behavior for resnet benchmark (#4231) David HouandGitHub 2024-04-19 17:07:46 -07:00
• f6eea03749 SAVE_SCHEDULE as contextvar (#4230) David HouandGitHub 2024-04-19 15:51:57 -07:00
• 2094b3b327 graph ScheduleItems (#4224) 3 people 2024-04-19 15:17:11 +03:00
• cd88afc98b datasets isn't a feature + filter docstrings (#4228) George HotzandGitHub 2024-04-19 16:16:10 +04:00
• b9570d6100 clean up update stats (#4226) George HotzandGitHub 2024-04-19 15:41:30 +04:00
• 1c87e5dbf6 fuzz schedule context vars (#4223) qazalandGitHub 2024-04-19 13:16:25 +03:00
• d99b512084 llm.c timing (#4219) George HotzandGitHub 2024-04-19 12:43:21 +04:00
• 43841a32b7 Merge pull request #4222 from Qazalin/fuzz-multi0 qazalandGitHub 2024-04-19 08:07:45 +03:00
• b2fe3884fc Merge branch 'master' into fuzz-multi0 qazalandGitHub 2024-04-19 07:56:26 +03:00
• abb10c83cd tunable multi output fusion qazal 2024-04-19 07:18:21 +03:00
• a1133beb80 KFD GEMM (#4221) chenyuandGitHub 2024-04-19 00:43:18 -04:00
• 3f3af0fb85 test_linearizer_failures 29 passes now (#4215) chenyuandGitHub 2024-04-18 19:49:23 -04:00
• 2ecd61e3e2 monkey patching (#4214) Elias WahlandGitHub 2024-04-19 01:20:52 +02:00
• 126826afc8 linearizer: refactor to define accs with potentially TC-modified idxs (#4211) Francis LamandGitHub 2024-04-18 12:31:06 -07:00
• 39b60a25f0 more llm c work (#4207) George HotzandGitHub 2024-04-18 22:20:44 +04:00
• f7416916df update resnet hparams based on BS=1632 RCP (#4210) chenyuandGitHub 2024-04-18 12:01:46 -04:00
• fa57c3e7ce continue llm.c (#4190) George HotzandGitHub 2024-04-18 10:57:54 +04:00
• 269a58d5fa tolist to return multidimensional list (#4192) geohotstanandGitHub 2024-04-18 11:43:10 +08:00
• 3644077a42 [MLPerf][UNet3D] Add DICE loss + metrics (#4204) 3 people 2024-04-17 20:09:33 -04:00
• cd801a15f3 scipy.signal.gaussian -> scipy.signal.windows.gaussian (#4205) chenyuandGitHub 2024-04-17 19:15:37 -04:00
• 6eef8ee22a Wikipedia download script for MLPerf BERT training (#4202) Elias WahlandGitHub 2024-04-17 22:34:57 +02:00
• f75020a903 minimal diff for multioutput reduce pairs (#4030) 3 people 2024-04-17 17:55:44 +03:00
• 8564e28a1b new memory scheduler with explicit refcounts (#4198) George HotzandGitHub 2024-04-17 08:46:47 +04:00
• c91b7b1739 test: add fuzz_matmul and better debugging for simple_matmul (#4199) Francis LamandGitHub 2024-04-16 20:40:31 -07:00
• ba8602612b Fuzz all permutations of schedule (#4136) 3 people 2024-04-17 04:03:21 +03:00
• 4ed6b42a8a fix kernargs check in kfd (#4194) nimlgenandGitHub 2024-04-17 00:44:50 +03:00
• 97d846dd67 in forced_realize, unchase last op if it is upcast (#4185) David HouandGitHub 2024-04-16 14:15:17 -07:00
• e9c1616b27 logging: change LOGKERN to LOGKERNS to match LOGOPS (#4193) Francis LamandGitHub 2024-04-16 13:08:32 -07:00
• 7fb220a567 touchup resnet_layer_bench (#4191) David HouandGitHub 2024-04-16 11:43:00 -07:00
• 1dbf3b2b19 Benchmarks for individual resnet layers (#4182) David HouandGitHub 2024-04-16 10:53:18 -07:00
• d49d4324a3 update docs (#4189) George HotzandGitHub 2024-04-16 16:07:02 +04:00
• 55ae73e951 Replicate llm.c in tinygrad (#4179) 4 people 2024-04-16 15:40:48 +04:00
• b6e7243bfa hotfix: skip slow pre-commit test geohot 2024-04-16 11:48:43 +04:00
• cda0010020 hotfix: docs-legacy geohot 2024-04-16 11:06:56 +04:00
• 8f749ae0eb New docs are in mkdocs (#4178) George HotzandGitHub 2024-04-16 10:59:51 +04:00
• aa093efa43 fix handcode_resnet50_opt flops count (#4184) chenyuandGitHub 2024-04-15 22:13:45 -04:00
• d5b67c1ca3 log resnet TRAIN_BEAM / EVAL_BEAM (#4181) chenyuandGitHub 2024-04-15 19:29:08 -04:00
• 9d2273235c search: BEAM_UOPS_MAX to prune candidates with too many uops (#4088) Francis LamandGitHub 2024-04-15 15:56:22 -07:00
• 286ea697f3 keep order in realizes (#4180) qazalandGitHub 2024-04-16 00:25:50 +03:00
• e14a9bca0c hotfix: bump line count to 7500 for NV backend geohot 2024-04-15 23:18:46 +04:00
• 6a2168e698 TRAIN_BEAM and EVAL_BEAM for resnet (#4177) chenyuandGitHub 2024-04-15 14:57:21 -04:00
• 4592fc8fe7 Multireduce Kernels - prereq refactor (#4173) TimmyandGitHub 2024-04-14 20:16:54 -04:00
• 593c90d7d6 Resnet fp16 training with fp32 master weight copy (#4144) 3 people 2024-04-14 08:25:08 -07:00
• e20d6f9221 correct resnet estimate time (#4169) chenyuandGitHub 2024-04-14 02:21:46 -04:00
• ea18d28253 some overview docs geohot 2024-04-13 17:01:09 -07:00
• 50e780a588 multitensor shouldn't recompile (#4164) George HotzandGitHub 2024-04-13 00:03:48 -07:00
• 599eb266b1 optionally use a copy kernel instead of SDMA (#4116) George HotzandGitHub 2024-04-12 23:10:41 -07:00
• ba7314c26b cleanup lbs (#4163) George HotzandGitHub 2024-04-12 22:32:16 -07:00
• a7c6864260 remove CAST_BEFORE_VIEW (#4152) chenyuandGitHub 2024-04-13 01:05:08 -04:00
• ebc94c9d6c rewrite the jit in the context of new schedule (#4162) George HotzandGitHub 2024-04-12 21:54:36 -07:00
• b67f759780 abstractions3 is currently wishful thinking (#4124) George HotzandGitHub 2024-04-12 16:46:01 -07:00
• 27a98aaecc Rewritten SVG Logos (#4150) MaximilianEmelandGitHub 2024-04-12 17:09:57 -04:00
• 63eb0a68af fix return dtype of gather (#4159) chenyuandGitHub 2024-04-12 16:25:12 -04:00
• d9c5a2b1bb fix return dtype of getitem Tensor indexing (#4158) chenyuandGitHub 2024-04-12 15:55:02 -04:00
• f6c8032e5d assert if expr_idxs return might be outside of int32 (#4157) chenyuandGitHub 2024-04-12 14:18:35 -04:00
• 24a27a01a9 hotfix: CUDA_P2P works (#4155) nimlgenandGitHub 2024-04-12 18:20:12 +03:00
• 5a57b48134 cuda p2p enable when available (#4153) nimlgenandGitHub 2024-04-12 16:21:54 +03:00
• 380f27d629 move sum acc_dtype into lazy so it applies to backward (#4149) chenyuandGitHub 2024-04-11 14:43:56 -04:00